If the 1981 text set the direction, here I try to give the method a clearer form: design as a process of refinement between abstract machines, with explicit relations and with a framework that allows verification and optimization.
The starting point is the idea that design should not be described only “narratively”, but captured in a model formal enough to make it possible to speak about steps, correctness and consequences. In this model, the abstract machine functions as the basic object: a mathematical description strict enough to be compared, refined and connected.
The aim is not to replace the designer, but to provide a discipline: a succession of transformations in which properties are explicitly followed. Here the ideas of intermediate form and module network also take shape, that is, a design that has both a vertical dimension (levels of abstraction) and a horizontal one (structure).
In the sequence of the decade, the 1983 article is a connecting link: it pushes the method toward a stable form, preparing the possibility of its instrumentation (Tokyo 1985) and, ultimately, of an INTERFORM-type system.